When are Elastic performance reviews? (2026)
Elastic is distributed-first with an open-source heritage and a search, observability and security product split. Cycle dates vary, so keep a written record that works asynchronously.
Once a year
No specific review month is published for this company, because none could be sourced from the company's own material. What follows is the cycle shape, which is what is actually knowable. Your own manager is the authority on this year's dates.
How the cycle works
Elastic has been distributed across many countries for most of its existence, and that is not a policy so much as the shape of the company. The consequence for evaluation is the same one that applies at any genuinely distributed organization: your contribution exists to the extent it was written down. There is no incidental visibility to fall back on.
The product has grown into three fairly distinct areas, search, observability and security, built on shared underlying technology. Which one you work in changes what evidence looks like considerably, and someone whose work spans two of them is doing something structurally valuable because the shared foundation is where the leverage is.
The open-source heritage also means a portion of the work is public. Public contributions, issue discussions and release notes are a durable, externally verifiable record, which is a preparation advantage worth using.
Cycle timing varies. Confirm with your manager.
What actually gets weighed
Search relevance and query performance are the foundational axes: result quality, latency at the tail, and index efficiency, all measured against realistic corpora rather than toy data.
Operational scale is the second. Customers run very large clusters, and cluster stability, shard behavior under pressure and recovery from node loss are the things that separate a system that works in a demo from one that works in production.
Cross-solution leverage is the third. Work in the shared foundation that improved all three product areas at once is worth more than the equivalent effort in one, and it should be reported as the aggregate.
Asynchronous collaboration quality is a real fourth axis in a company this distributed: someone whose written proposals let colleagues in other time zones act without a meeting has removed real friction.
What to have ready
- Relevance and latency results, including the tail, measured on realistic data.
- Cluster-scale behavior: recovery, rebalancing, stability under load.
- Foundation work reported as the sum of its effect across all three product areas.
- Public artifacts, since the open development model leaves an externally verifiable trail most people never use in their own review.
- Written proposals that unblocked people across time zones, with the specific colleagues named.